Implications of observed and simulated ambient flow in monitoring wells

A recent paper by Hutchins and Acree (2000) has called attention to ground water sampling bias due to ambient (natural gradient-induced) flows in monitoring wells. Data collected with borehole flowmeters have shown that such ambient flows are ubiquitous in both confined and unconfined aquifers. Developed herein is a detailed three-dimensional model of flow and transport in the vicinity of a fully penetrating monitoring well. The model was used to simulate a measured ambient flow distribution around a test well in a heterogeneous aquifer at the Savannah River Site (SRS) near Aiken, South Carolina. Simulated ambient flows agreed well with measurements. Natural flow was upward, so water entered the well mainly through high K layers in the lower portion of the aquifer and exited through similar layers in the upper portion. The maximum upward discharge in the well was about 0.28 L/min, which implied an induced exchange of 12 m(3)/month from the bottom half of the aquifer to the upper half. Tracer transport simulations then illustrated how a contaminant located initially in a lower portion of the aquifer was continuously transported into the upper portion and diluted throughout the entire well by in-flowing water. Even after full purging or micropurging, samples from such a well will yield misleading and ambiguous data concerning solute concentrations, location of a contaminant source, and plume geometry. For all of these reasons, use of long-screened monitoring wells should be phased out, unless an appropriate multilevel sampling device prevents vertical flow.